Learning-Based Optimal Entry Guidance with Path Constraints

This work is dedicated to generating optimal guidance commands for entry vehicles subject to path constraints. A multi-constrained nonlinear optimal control problem for maximizing the terminal speed of the entry vehicle is formulated and converted into a regularized equality-constrained one using the saturation function method. Subsequently, a parameterized system is established based on the necessary conditions derived from Pontryagin's Minimum Principle. This enables the efficient generation of optimal state-control samples, crucial for training Neural Networks (NNs) in the supervised learning framework. Finally, the performance of the learning-based entry guidance algorithm is numerically presented.